The subjective science of persona building by Stephen Masiclat - Syracuse University
Personas guide design and ensure services have a relevant constituency, but they have lacked a low-cost, scientifically valid method for genesis. This presentation shows how Q-Methodolgy defines rigorous personas to guide and test the service design process
SDNC13 -Day2- The subjective science of persona building by Stephen Masiclat
1.
2. service design research
building better personas to guide design
S.Masiclat The S.I.Newhouse School of Public Communications, Syracuse University
SM(D) strategic media x design
3. the question a movie studio wants to know if a
proposed new service has a market—
if we build it, will they come?
the proposed service is social+movie; is there an audience
who wants to be social (i.e. tweet,
text with friends, check in on
Foursquare, etc) while they are
watching a movie?
the initial method was to form focus groups wherein
we asked participants to react to
a customer story. . .
4. the initial method was to form focus groups wherein
we asked participants to react to
a customer story. . .
. . .this did not proceed as
anticipated. . .
in three attempts we never got past
the initial phase of the customer story:
“Imagine you’ve arrived at the
multiplex
to watch a movie. It’s one
you’ve talked
about with friends of yours, but
they
couldn’t make it with you to the
theater
this time. One of the options for
the
5.
6. starting over . . . is there a method that can:
validate market sectors for an as yet
undesigned service and define the
personas that inhabit those sectors
protect the privacy of the research
subjects
We did a literature review to see if
anyone had published previous research in generating
personas to guide design using a method that met the criteria
above. . .
One research finding stood
out.
7. ‘Personas are
bollocks*.’
“The most serious limitation of the Personas method is that
it is difficult or impossible to verify that personas are
accurate. This involves several aspects: a problematic
relationship between personas and user populations;
burdens on inference related to personas’ high specificity;
and the possibility that personas are non-falsifiable.”
Chapman, C.N, and Milham, R. P (2006) 'The persona's new clothes: methodological and practical
arguments against a popular method' Proceedings of the Human Factors and Ergonomics Society 50th
Annual Meeting, pp. 634 –636. Available online:
http://cnchapman.files.wordpress.com/2007/03/chapman-milham-personas-hfes2006-0139-0330.pdf
8. Q-methodology can address these issues
Q is designed to study subjectivity scientifically:
it captures a personal point of view
it is arises from an internal frame of reference
Q measures holistic attitudes, not bits and pieces
overcomes inarticulateness by providing statements for sorting
9. Q-methodology was developed by
William Stephenson
PhD in physics (1926) and psychology (1929). Stephenson’s work
in psychometrics advanced his advisor Charles Spearman’s work
in factor analysis by applying eigenvector rotational maths to
provide alternate views of data.
http://en.wikipedia.org/wiki/File:Stephenson.jpg
Q analysis allows you to conduct intensive data analysis across
multiple cases while respecting the captured contours of the data.
10. Q-Method steps: the Q-Sample
Develop a concourse—a comprehensive set of statements—
that capture the thoughts and opinions around a topic. For our
study, these were sometimes speculative.
From the concourse, select a representative Q-Sample of 3540 statements.
35 | I miss watching movies
with my distant friends and family.
2 | “Checking in” to a show with
an app makes watching TV a chore.
15 | I’m willing to see a movie
I’m not interested in if I see it
with friends.
18 | Using a mobile phone in
a movie theater is rude!
20 | I enjoy being able to watch a
movie whilst using my computer
19 | Watching a movie on a big
screen is much better than on a
tablet or mobile phone.
30 | People who tweet about
the TV and movies they watch
are narcissists.
12 | My family and I watch some
TV shows ‘together’ even if we’re in
different places.
6 | It ruins the movie experience
when people talk or text in
the theater.
11 | Sometimes, a moment in a
movie makes me so excited that
I have to tweet or share it.
1 | I like perfect silence when I’m
watching a movie in a theater.
4 | My children like talking with their
friends at the movie theater.
38 | Websites already spy on me, I
don’t want my phone to do the same
thing.
17 | My friends’ comments can
make old movies new again.
34 | When I’m captivated by a
movie, the last thing on my mind
is sharing with my social network.
11. Q-Method steps: the preliminary Q-Sort
At this stage, subjects sort the statements into three piles according
to a condition of instruction. For our study, we asked people to sort the
items from those they most agreed with, to those with which they most
disagreed. Statements that do not elicit a strong opinion are left in the
middle, “unsure” area.
generally disagree
generally agree
leaning, but not sure
34
22
10
6
13
16
30
9
18
8
2
7
27
4
26
21
3
11
23
31
14
1
38
19
17
20
15
5
36
24
29
32
28
25
12
33
35
37
12. Q-Method steps: the final Q-Sort
Now, subjects move to the positive pile, find the statements they are
most strongly aligned to, and place those in the +4 category. Then
they move to the negative pile, find the statements they most
negatively react to, and place those in the -4 category. They repeat
the process moving back and forth on the scale.
-4
-3
-2 -1
+1 +2
0
+3
+4
5
22
2
24
12
6
30
15
18
14
10
23
27
32
38
11
3
21
31
7
1
16
19
25
28
34
33
29
4
9
8
13
35
20
17
36
26
37
13. Q-Method steps: factor analysis
Using low-cost (and free) software Q-Sorts are inter-correlated and
plotted. Individual plots are then rotated (the essential factor analysis
technique). Researchers next apply rotation transformations
(eigenvector varimax) to discern the factors.
14. Factors emerge and they describe a
group’s shared ideas
Q-Sorts that resemble each other closely coalesce into groups that
share common attitudes.
The method always seeks factor solutions that are most different from
each other. Generally, only a few factor solutions emerge as the most
distinct, most representative descriptions of the topic under study.
The factors are the personas. Combined with traditional
demographic data, you get a statistically valid descriptor of the
different ideas people hold, and a beginning description of the
people themselves.*
15. This is not a maths lecture. . .
The keys for service design researchers are:
This is a rigorous (scientific, falsifiable, repeatable, statistically
valid, etc.) method that automatically generates personas based
on a wholistic, shared mind-set.
The findings emerge from the data. There is no a priori persona
to which we try to align our services architecture, and around
which we base our research.
This method is low-cost; it is fast, you don’t need very large
sample sizes, and the analysis tools are free or cheap.
Most classically-educated executives are used to seeing largescale user data, or survey results from samples chosen for
confidence interval. Therefore, you might still be asked to
defend these results. . . mathematically.
16. The results are not random.
In our movie service study, we had 38 statements. That means
there are 38! (38 factorial) ways to arrange the statements.
How big is 38 factorial ?
number of ways to arrange
divided by 9!
523,022,617,466,601,111,760,007,224,100,074,291,200,00
0,000
14,41,310,123,089,178,548,721,360,295,690,240,000,000
17. The results are not random.
In our movie service study, we had 38 statements. That means
there are 38! (38 factorial) ways to arrange the statements.
How big is 38 factorial ?
number of ways to arrange
divided by 9!
divided by 7,000,000,000
523,022,617,466,601,111,760,007,224,100,074,291,200,00
0,000
14,41,310,123,089,178,548,721,360,295,690,240,000,000
205,901,446,155,596,935,531,622,899,384
205 octillion, 901 septillion, 446 sextillion, 155 quintillion,
596 quadrillion, 935 trillion, 531 billion, 622 million, 899
thousand, 384 planet Earths to duplicate these results from
random processes.
18. Our study yielded 3 factors—three
personas with clear ideas about, and
strong attitudes toward this proposed
service.
The vast majority (we estimate 70% of the general population) who
demand absolute darkness and silence. The movie experience is about
immersion, and any distraction is a diminution of the experience.
Bros. A predominantly male group interested in social media-enabled
connection around sports. For them, social media apps could preserve
their friendly rivalries and increase the value of sports media.
Young women who absolutely positively want the movie+social experience.
Now. Like, seriously, where is this movie playing?
19. Our study done in June 2012.
In September 2013, Disney announces their Second Screen Experience.
images by Daniel Nasserian | http://www.disneygeekery.com
20. Design of the service proceeds from a
clearly defined and validated persona
images by Daniel Nasserian | http://www.disneygeekery.com
21. Thank you.
Free Q-Method analysis software: http://schmolck.userweb.mwn.de/qmethod/
International Society for the Scientific Study of Subjectivity (ISSSS) http://qmethod.org/
Big number calculations by Wolfram Alpha
Stephen Masiclat
Director, Graduate Program in New Media Management
The S.I.Newhouse School of Public Communications
Syracuse University
SM(D)
strategic media x design
@masiclat